Version: HEAD (using KDE KDE 3.5.0) Installed from: Compiled From Sources OS: Linux An LFI requirement reads: "It shall be possible to calculate the white noise limit from an input scientific data stream (where the input data will allow such estimation. For example it is expected that for LFI data this will be possible only for data in differenced mode)"
The LFI team would also like to calculate the knee frequency of a power spectrum. I would like to implement this as a fitting plugin. The plugin will operate on a power spectrum, fitting for the 1/(f^a) noise and the white noise limit at high frequencies. The options will be something like.. - # of points to fit 1/f noise (or a maximum 1/f noise frequency). - # of points to fit white noise (or a minimum white noise frequency). It will produce scalars for the knee frequency, the white noise limit, and the fit parameter a. It will produce a plot with the original power spectrum, a fit curve which shows the intersection of the white noise limit with the 1/f noise, and labels containing the values of the calculated scalars. Does this sound good?
Sounds good, but this should be against bug report 115451 and not this one.
*** This bug has been marked as a duplicate of 115451 ***